Chi-Shih Jao, Yusheng Wang, Yu-Wei Lin, and Andrei M. Shkel, University of California, Irvine

View Abstract Sign in for premium content

Abstract:

In this paper, we propose a hybrid altimeter for foot-mounted Inertial Navigation Systems (INS) aided by a barometer and a downward-facing ultrasonic sensor. The goal of developing the hybrid altimeter was to minimize the usage of barometric altimeter in height estimation as measurements of barometers are affected by changes in ambient temperature and air pressure. In this paper, we show that a shoe-mounted downward-facing ultrasonic sensor alone could be used as an altimeter on a variety of terrains, from flat surfaces to uneven stairs, by simultaneously estimating shoe height and floor elevation. We will refer to this type of altimeter as the ultrasonic altimeter. To account for other common indoor terrains, such as ramps and elevators, the developed hybrid altimeter uses a Multi-Model Kalman Filter to fuse measurements of a barometer and the ultrasonic altimeter. In the fusion process, the hybrid altimeter adaptively selects weights for the barometric and ultrasonic measurements based on the terrains under operation. We utilized a foot-mounted Inertial Measurement Unit (IMU) to realize the detection of elevators and ramps. We conducted three series of experiments to test the performance of the proposed hybrid altimeter. The first experiment showed that the proposed altimeter is less sensitive to temperature and air pressure changes in the surrounding environment than a barometer. The second series of experiments investigated the performance of the hybrid altimeter in the case of walking slowly on a flat plane, and the corresponding experimental results indicated that the Zero velocity UPdaTe (ZUPT)-based INS aided by the hybrid altimeter improved the Root Mean Square Error (RMSE) along the vertical direction by 96%, as compared to the standalone ZUPT-based INS, and by 97%, when compared to the ZUPT-based INS aided by the barometer. In the third series of experiments, we studied the navigation accuracy of the hybrid altimeter for common indoor terrains, such as flat surfaces, stairs, ramps, and elevators. The results of this series of experiments showed that the RMSE of the ZUPT-based INS aided by the hybrid altimeter outperformed the RMSE of the standalone ZUPT-based INS by 91%. When compared to the RMSE of the ZUPT-based INS aided by the barometer vs. the hybrid altimeter, in the latter case the accuracy was increased by 41% (0.15 ????). The primary residual error sources, in our opinion, contributing to the vertical errors of the ZUPT-based INS aided by the hybrid altimeter are the underestimate of stair height and the false alarm of stair detection of the ultrasonic altimeter.